Bus scan is responsible for finding devices over *all* buses.
Some of these buses might not be able to scan but that should
not prevent other buses to be scanned.
Same is the case for probing. It is possible that some devices which
were scanned didn't have a specific driver. That should not prevent
other buses from being probed.
Signed-off-by: Shreyansh Jain <shreyansh.jain@nxp.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
This function can be used to check the role of a specific lcore.
Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
GCC does have the __get_cpuid_count builtin which checks for maximum
supported leaf, but implementations differ between CLANG and GCC.
This change provides an implementation compatible with both GCC and
CLANG 3.4+.
Signed-off-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Split pci_vfio_map_resource for primary and secondary processes.
Save all relevant mapping data in primary process to allow
the secondary process to perform mappings.
Signed-off-by: Jonas Pfefferle <jpf@zurich.ibm.com>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
DMA window size needs to be big enough to span all memory segment's
physical addresses. We do not need multiple levels of IOMMU tables
as we already span ~70TB of physical memory with 16MB hugepages.
Signed-off-by: Jonas Pfefferle <jpf@zurich.ibm.com>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Normally, command line argument strings are considered immutable, but
SPDK [1] and urdma [2] construct argv arrays to pass to rte_eal_init().
These strings are allocated using malloc() and freed after DPDK
initialization with free(). However, in the case of --file-prefix and
--huge-dir, DPDK takes the pointer to these strings in argv directly. If
a secondary process calls rte_eal_pci_probe() after rte_eal_init()
returns, as is done by SPDK, this causes a use-after-free error because
the strings have been freed by the calling code immediately after
rte_eal_init() returns.
This problem was observed when running SPDK example programs as a
secondary process and causes the secondary processes to fail:
Starting DPDK 16.11.1 initialization...
[ DPDK EAL parameters: identify -c 4 --file-prefix=spdk3260 --base-virtaddr=0x1000000000 --proc-type=auto ]
EAL: Detected 40 lcore(s)
EAL: Auto-detected process type: SECONDARY
EAL: Probing VFIO support...
EAL: VFIO support initialized
EAL: PCI device 0000:81:00.0 on NUMA socket 1
EAL: probe driver: 8086:953 spdk_nvme
EAL: cannot connect to primary process!
EAL: Error - exiting with code: 1
Cause: Requested device 0000:81:00.0 cannot be used
Running strace shows that the file prefix has been zero'd out by the
time that the secondary process attempts to probe the NVMe device.
The use-after-free errors can be easily detected with valgrind:
==8489== Invalid read of size 1
==8489== at 0x4C30D22: strlen (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==8489== by 0x58DB955: vfprintf (vfprintf.c:1637)
==8489== by 0x59A4685: __vsnprintf_chk (vsnprintf_chk.c:63)
==8489== by 0x59A45E7: __snprintf_chk (snprintf_chk.c:34)
==8489== by 0x1246AB: get_socket_path.constprop.0 (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x124B09: vfio_mp_sync_connect_to_primary (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x123BE4: vfio_get_group_fd.part.1 (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x124366: vfio_setup_device (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x126C8A: pci_vfio_map_resource (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x12B115: pci_probe_all_drivers.part.0 (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x12B596: rte_eal_pci_probe (in /home/pmacarth/src/spdk/examples/nvme/identify/identify)
==8489== by 0x11D5B5: spdk_pci_enumerate (pci.c:147)
==8489== Address 0x63f362e is 14 bytes inside a block of size 32 free'd
==8489== at 0x4C2ED5B: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==8489== by 0x11E6FB: spdk_free_args (init.c:136)
==8489== by 0x11EBF5: spdk_env_init (init.c:309)
==8489== by 0x10D2AA: main (identify.c:976)
==8489== Block was alloc'd at
==8489== at 0x4C2DB2F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==8489== by 0x11E7D7: _sprintf_alloc (init.c:76)
==8489== by 0x11EA78: spdk_build_eal_cmdline (init.c:251)
==8489== by 0x11EA78: spdk_env_init (init.c:282)
==8489== by 0x10D2AA: main (identify.c:976)
==8489==
Fix this by using strdup() to create separate memory buffers for these
strings. Note that this patch will cause valgrind to report memory
leaks of these buffers as there is nowhere to free them. Using static
buffers is an option but would make these strings have a fixed maximum
length whereas there is currently no limit defined by the API.
[1] http://spdk.io
[2] https://github.com/zrlio/urdma
Fixes: af75078fece3 ("first public release")
Cc: stable@dpdk.org
Signed-off-by: Patrick MacArthur <patrick@patrickmacarthur.net>
Acked-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
If mmap fails, it will return the value MAP_FAILED. Checking for this
return code allows us to properly identify mmap failures and report
them as such to the calling function.
Signed-off-by: Seth Howell <seth.howell@intel.com>
Acked-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
malloc_elem_free() is clearing(setting to 0) the trailer cookie when
RTE_MALLOC_DEBUG is enabled. In case of joining free neighbor element,
part of joined memory is not getting cleared due to missing the length
of trailer cookie in the middle.
This patch fixes calculation of free memory length to be cleared in
malloc_elem_free() by including trailer cookie.
Fixes: af75078fece3 ("first public release")
Cc: stable@dpdk.org
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Acked-by: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
Currently, enabling assertion have to set CONFIG_RTE_LOG_LEVEL to
RTE_LOG_DEBUG. CONFIG_RTE_LOG_LEVEL is the default log level of control
path, RTE_LOG_DP_LEVEL is the log level of data path. It's a little bit
hard to understand literally that assertion is decided by control path
LOG_LEVEL, especially assertion used on data path.
On the other hand, DPDK need an assertion enabling switch w/o impacting
log output level, assuming "--log-level" not specified.
Assertion is an important API to balance DPDK high performance and
robustness. To promote assertion usage, it's valuable to unhide
assertion out of COFNIG_RTE_LOG_LEVEL.
In one word, log is log, assertion is assertion, debug is hot pot :)
Rationale of this patch is to introduce an dedicate switch of
assertion: RTE_ENABLE_ASSERT
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Acked-by: Gaetan Rivet <gaetan.rivet@6wind.com>
We remove xen-specific code in EAL, including the option --xen-dom0,
memory initialization code, compiling dependency, etc.
Related documents are removed or updated, and bump the eal library
version.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Previously, to get MFN address in dom0, this API is a wrapper to
obtain the "physical address".
As we will removed xen dom0 support, this API is not necessary.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
IGB_UIO compilation recently got enabled for ARM64 by default
The igb_uio compilation against ARM64 based stock 4.x (e.g. 4.13)
kernel is giving compilation warnings:
igb_uio.c: In function ‘igbuio_pci_irqcontrol’:
igb_uio.c:115:25: error: implicit declaration of function
‘irq_get_irq_dat ’ [-Werror=implicit-function-declaration]
struct irq_data *irq = irq_get_irq_data(udev->info.irq);
^
igb_uio.c:115:25: error: initialization makes pointer from integer without
a cast [-Werror=int-conversion]
Fixes: d196343a258e ("igb_uio: use kernel functions for masking MSI-X")
Signed-off-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Tested-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
This is not bugfix, but it's convenient to help developer
to review and maintain the igbuio codes.
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
This patch adds MSI IRQ mode in a way, that should
also work on older kernel versions. The base for my patch
was an attempt to do this in cf705bc36c which was later
reverted in d8ee82745a. Compilation was tested on Linux 3.2,
4.10 and 4.12.
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
For better readability throughout the module, the destruction
order is changed to the exact inverse construction order.
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
The patch which introduced the usage of pci_alloc_irq_vectors
came after the patch which switched to non-threaded ISR (f0d1896fa1),
but did not use non-threaded ISR, if pci_alloc_irq_vectors
is used.
Fixes: 99bb58f3adc7 ("igb_uio: switch to new irq function for MSI-X")
Cc: stable@dpdk.org
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
igb_uio already allocates irqs using pci_alloc_irq_vectors on
recent kernels >= 4.8. The interrupt disable code was not
using the corresponding pci_free_irq_vectors, but the also
deprecated pci_disable_msix, before this fix.
Fixes: 99bb58f3adc7 ("igb_uio: switch to new irq function for MSI-X")
Cc: stable@dpdk.org
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
Interrupt setup code in igb_uio has to deal with multiple
types of interrupts and kernel versions. This patch moves
the setup and teardown code into own functions, to make
it more readable.
Signed-off-by: Markus Theil <markus.theil@tu-ilmenau.de>
Tested-by: Markus Theil <markus.theil@tu-ilmenau.de>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
detect SLE version reverse chronologically as ">=" is being used.
Fixes: 2972254ce163 ("kni: fix build on Suse 12 SP3")
Cc: stable@dpdk.org
Signed-off-by: Nirmoy Das <ndas@suse.de>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
DPDK has support for both sw and hw mempool and
currently user is limited to use ring_mp_mc pool.
In case user want to use other pool handle,
need to update config RTE_MEMPOOL_OPS_DEFAULT, then
build and run with desired pool handle.
Introducing eal option to override default pool handle.
Now user can override the RTE_MEMPOOL_OPS_DEFAULT by passing
pool handle to eal `--mbuf-pool-ops-name=""`.
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
iova autodetection depends on rte_bus_scan result. Result of bus scan will
have updated device_list and each device in that list has its '.kdev' state
updated. That kdrv state used to detect iova mapping mode for that device.
_device_parse() has dependency on rt_bus_scan so,
Below calls moved up in the eal initialization order:
- eal_option_device_parse
- rte_bus_scan
And based on the result of rte_bus_scan_iommu_class - select iova
mapping mode.
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Introducing rte_eal_iova_mode() helper API. This API
used by non-eal library for detecting iova mode.
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
API(rte_bus_get_iommu_class) helps to automatically detect and select
appropriate iova mapping scheme for iommu capable device on that bus.
Algorithm for iova scheme selection for bus:
0. Iterate through bus_list.
1. Collect each bus iova mode value and update into 'mode' var.
2. Mode selection scheme is:
if mode == 0 then iova mode is _pa,
if mode == 1 then iova mode is _pa,
if mode == 2 then iova mode is _va,
if mode == 3 then iova mode ia _pa.
So mode !=2 will be default iova mode (_pa).
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Get iommu class of PCI device on the bus and returns preferred iova
mapping mode for that bus.
Patch also introduces RTE_PCI_DRV_IOVA_AS_VA drv flag.
Flag used when driver needs to operate in iova=va mode.
Algorithm for iova scheme selection for PCI bus:
0. If no device bound then return with RTE_IOVA_DC mapping mode,
else goto 1).
1. Look for device attached to vfio kdrv and has .drv_flag set
to RTE_PCI_DRV_IOVA_AS_VA.
2. Look for any device attached to UIO class of driver.
3. Check for vfio-noiommu mode enabled.
If 2) & 3) is false and 1) is true then select
mapping scheme as RTE_IOVA_VA. Otherwise use default
mapping scheme (RTE_IOVA_PA).
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Introducing rte_pci_get_iommu_class API which helps to get iommu class
of PCI device on the bus and returns preferred iova mapping mode for
PCI bus.
Patch also adds rte_pci_get_iommu_class definition for:
- bsdapp: api returns default iova mode.
- linuxapp: Has stub implementation, Followup patch has complete
implementation.
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
Export rte_pci_match() function as it needed in the followup patch.
Signed-off-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: Hemant Agrawal <hemant.agrawal@nxp.com>
The compilation with gcc-6.3.0 and EXTRA_CFLAGS=-Og gives the following
error:
CC eal_pci_uio.o
eal_pci_uio.c: In function ‘pci_get_uio_de ’:
eal_pci_uio.c:221:9: error: ‘uio_num’ may be used uninitialized in
this function [-Werror=maybe-uninitialized]
return uio_num;
^~~~~~~
This is a false positive: gcc is not able to see that when e != NULL,
uio_num is always set. Fix the warning by initializing uio_num to -1,
and by the way, change the type to int.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
The assertion of return value from the open() function is done against
0, while it is a correct value - open() returns -1 in case of an error.
It causes problems while trying to run as a daemon, in which case, this
call to open() will return 0 as a valid descriptor.
Fixes: b94e5c9406b5 ("eal/arm: add CPU flags for ARMv7")
Fixes: 97523f822ba9 ("eal/arm: add CPU flags for ARMv8")
Fixes: 9ae155385686 ("eal/ppc: cpu flag checks for IBM Power")
Cc: stable@dpdk.org
Signed-off-by: Lukasz Majczak <lma@semihalf.com>
Acked-by: Jan Viktorin <viktorin@rehivetech.com>
Acked-by: Chao Zhu <chaozhu@linux.vnet.ibm.com>
When compiled on Ubuntu with extra warnings enabled, the rte_strerror()
function triggered a warning about an unused return value from
strerror_r(). Rather than always have this warning disabled, we fix this,
and in the process do some cleanup of the code so as to reduce the
complexity of the fix, e.g. not having the #ifdef macros inside the
snprintf call.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
Any log messages during bus initialization maybe lost because
the bus registration constructor is called before the logging constructor.
Fixes: a97725791eec ("bus: introduce bus abstraction")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
It is a reminder that the constructors without priority
get the lowest priority.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Some compilers detect this error:
error: ‘ids[0]’ may be used uninitialized in this function
ret = rte_service_map_lcore_set(i, ids[lcore_iter], 1);
It can be reproduced very easily on Fedora 21 with
gcc-4.9.2-6.fc21.x86_64.
Fixes: 21698354c832 ("service: introduce service cores concept")
Cc: stable@dpdk.org
Signed-off-by: Yi Yang <yi.y.yang@intel.com>
Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
lcore_states store the state of the lcore. Fixing the invalid
dereference of lcore_states with service number
Unit test case service_lcore_start_stop fails with the above fix.
Service core was stopped without stopping the service.
This commit fixes the test by adding negative and positive cases of
stopping the service lcore before and after stopping the service
respectively
Fixes: 21698354c832 ("service: introduce service cores concept")
Fixes: f038a81e1c56 ("service: add unit tests")
Signed-off-by: Guduri Prathyusha <gprathyusha@caviumnetworks.com>
Reviewed-by: Harry van Haaren <harry.van.haaren@intel.com>
This commit adds a section to the service register function
to make it clear that registering a service, must not configure
service-cores (eg: adding lcores or changing mappings).
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit adds a new flag that the component (or "backend")
can use to indicate readyness. The service function callback
will not be called until the component sets itself as ready.
The use-case behind adding this feature is eg: a service that
requires configuration before it can start. Any service that
emulates an ethdev will have rte_eth_dev_configure() called,
and only after that the service will know how many queues/etc
to allocate. Once that configuration is complete, the service
marks itself as ready using rte_service_component_runstate_set().
This feature request results from prototyping services, and
requiring a flag in each service to note "internal" readyness.
Instead that logic is now lifted to the service library.
The unit tests have been updated to test the component runstate.
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This aligns with the service stats, which are currently
also reset on read. A generic statistics API would be
helpful for the service library in future.
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit ensures that after an lcore is added, that
it is in the WAIT state. Previously, adding an lcore did
not ensure that the core was ready for being relaunch, which
would cause errors during lcore_start(). Now that the lcore is
ensured to be in WAIT state by the lcore_add() function, this
is no longer an issue.
Fixes: 21698354c832 ("service: introduce service cores concept")
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Previously to this commit, the return value of the following
functions was the mask value, instead of a neat 0 or 1.
This commit fixes this using the "!!" trick, to force the
number to 1 instead of the bitmask value itself, bringing
the return value in line with the function documentation.
Fixes: 21698354c832 ("service: introduce service cores concept")
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Services can be registered and unregistered, and "holes" can
appear in the contiguous array of services if a service is
unregistered. As a result, we must never iterate to the
number of services (as counted by rte_service_count), instead
scanning the service array and checking if the service is valid.
After this commit, the rte_service_count variable is only used
for its intended purpose; tracking the number of services that
are present.
Fixes: 21698354c832 ("service: introduce service cores concept")
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit fixes an issue in the service runner function,
where the atomic value was not cleared on exiting the service
function. This resulted in future attempts to run the service
to appear like the function was running, however it was in
reality deadlocked.
This commit refactors the atomic handling to be more readable,
by splitting the implementation code into a new static inline
function. The remaining flow control of atomics in the existing
function is refactored for readability.
Fixes: 21698354c832 ("service: introduce service cores concept")
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit reworks the service_get_by_name() function to
accept an integer, and removes the service_get_by_id() function.
All functions now accept an integer argument representing the
service, so it is no longer required to expose the service_spec
pointers to the application.
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit reworks the unregister API to accept an integer.
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit reworks the statistics functions to use integer ids
for services instead of pointers. Passing UINT32_MAX to the dump
function prints all info, similar to passing NULL previously.
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
This commit reworks the API to move from two separate start
and stop functions, to a "runstate" API which allows setting
the runstate. The is_running API is replaced with an function
to query the runstate. The runstate functions take a id value
for service. Unit tests and the eventdev sw pmd are updated.
Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com>
Acked-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
